A silo architecture can be a replicated hardware/software configuration. The application hosted on these silos can be a staged, event-driven application. A problem occurs if one of the stages in a silo hangs, and the load balancer, which is application agnostic, is not able to assess the situation and continues to the send the transactions to the disabled silo with the downed stage. The transactions will be processed by the disabled silo until they reach the downed stage. This situation will cause exponential degradation in the performance of the silo and the loss of the transactions being routed through the disabled silo. Eventually this disabled silo will have to be shut down and corrected; however, shutting down the silo will result in a loss of transactions that have already been fed by the load balancer to the silo. The loss of any transactions is not desirable and may result in complaints and dissatisfied customers.
Thus, there is an opportunity for systems and methods for dynamic transaction migration in an event-driven, multi-silo architecture.